Workshop date: December 14, 2026 (Monday) — Middleware 2026 Workshops Day 1
Workshop hours: 8:30–16:30 (CET), across the four workshop session slots
Location: Universitat Rovira i Virgili, Tarragona, Spain (hybrid — in-person and remote participants); room to be announced by the Middleware organizers
Hosted by: 27th ACM/IFIP International Middleware Conference (December 14–18, 2026) — see the full conference program
Accepted papers: 12, presented across four sessions, plus two keynotes
| Time (CET) | |
|---|---|
| 8:00–8:30 | Registration (shared, all workshop tracks) |
| 8:30–10:00 | WoAIS Session 1 — Opening & Keynote 1: Agentic AI Meets Serverless |
| 8:30–8:40 | Welcome, workshop overview and online polls (Aleksander Slominski and Pedro García Lopez) |
| 8:40–9:20 | Keynote 1: to be announced [35 min + 5 min Q&A] |
| 9:20–9:40 | Is Serverless a Misfit for Agentic AI? (Bita Asoodeh, University of Edinburgh) [16 min + 4 min Q&A] |
| 9:40–10:00 | Why “Succeeded” Is Not Enough: Four Axes for Measuring Production Agents (Girish Konda, Microsoft) [16 min + 4 min Q&A] |
| 10:00–10:30 | Coffee break (shared) |
| 10:30–12:00 | WoAIS Session 2 — LLM Workflows, Routing and Composition |
| 10:30–10:50 | Toward Spec-Driven Generation of Serverless Workflows with LLMs (Christian Chisté, University of Innsbruck) [16 min + 4 min Q&A] |
| 10:50–11:10 | Dependency-Aware Task Decomposition and Parallel Multi-LLM Execution on Serverless Workflows (Jackson Beem, Oakland University) [16 min + 4 min Q&A] |
| 11:10–11:30 | Static Before Smart: Exact Cost Accounting and Baselines for Pay-per-Token LLM Routing (Arjun Subramanian, MIT CSAIL) [16 min + 4 min Q&A] |
| 11:30–11:50 | Making Agentic Composition Measurable: Experience from a Kubernetes-Native Placement Backend (Gabriele Scaffidi Militone, Politecnico di Torino) [16 min + 4 min Q&A] |
| 11:50–12:00 | Open discussion across session-2 talks — or up to 2 demos (5 min each) |
| 11:50–11:55 | ↳ Demo slot 1 — available |
| 11:55–12:00 | ↳ Demo slot 2 — available |
| 12:00–13:00 | Lunch (shared) |
| 13:00–14:30 | WoAIS Session 3 — Serverless AI Workloads, Efficiency and Privacy |
| 13:00–13:20 | LLM Fine-Tuning Workloads: Characterization, Challenges, and Opportunities for Serverless AI (Radu Nicolae, Vrije Universiteit Amsterdam) [16 min + 4 min Q&A] |
| 13:20–13:40 | BlockGuard: Serverless Selective Homomorphic Encryption for Federated Learning (Amine Barrak, Oakland University) [16 min + 4 min Q&A] |
| 13:40–14:00 | Towards Energy-Efficient Serverless Computing with Hardware Isolation (Natalie Carl, Technische Universität Berlin) [16 min + 4 min Q&A] |
| 14:00–14:20 | How Much Do Co-Located Containers Really Share? A Pairwise Measurement Study (Hanlin Wu, KU Leuven) [16 min + 4 min Q&A] |
| 14:20–14:30 | Open discussion across session-3 talks — or up to 2 demos (5 min each) |
| 14:20–14:25 | ↳ Demo slot 3 — available |
| 14:25–14:30 | ↳ Demo slot 4 — available |
| 14:30–15:00 | Coffee break (shared) |
| 15:00–16:30 | WoAIS Session 4 — Serverless Platforms and Execution Semantics, with Keynote 2 |
| 15:00–15:20 | Towards Stateful Serverless Via Suspension: First Steps by AWS Lambda MicroVMs (Iván Molinero-Moreno, Universitat Rovira i Virgili) [16 min + 4 min Q&A] |
| 15:20–15:40 | AtomicFaaS: Offloading Side-Effect Compensation to the Serverless Platform (Jonas Heisterberg, Technische Universität Berlin) [16 min + 4 min Q&A] |
| 15:40–16:20 | Keynote 2: to be announced [35 min + 5 min Q&A] |
| 16:20–16:30 | Closing remarks, feedback link, thank you! (Workshop co-chairs) |
Paper titles link to the talks page for abstracts and the full author list. See the Call for Papers for the workshop scope.